From my experience, time of day does have an impact.
Not so much in terms of “between these two hours”, more in terms of evening as opposed to afternoon as opposed to morning.
Eliminate race type
- MemphisFlash
- Posts: 2128
- Joined: Fri May 16, 2014 10:12 pm
- Location: Leicester
Well that didn't go quite to plan. A Few tweaks here and there to stop the Rot
You do not have the required permissions to view the files attached to this post.
-
- Posts: 22
- Joined: Mon Sep 28, 2015 8:47 pm
Thanks for all of the comments
Been running this "mut" strategy for a 3 weeks now and still in the green (just), take the
points being made about times of races. Seems to me that to get a consistent and
meaningful profit I need to look into amount matched, track, time or meeting and quality
of meeting.
Will keep you posted
Been running this "mut" strategy for a 3 weeks now and still in the green (just), take the
points being made about times of races. Seems to me that to get a consistent and
meaningful profit I need to look into amount matched, track, time or meeting and quality
of meeting.
Will keep you posted
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
Although the tricky issue is time frame.
Memphis's was taken over a day - which is kinda mad
But if it was taken over say, 1/2 a yr, I think an 80 sample "could" be enough i.e. it's an extremely rare (but consistent) pattern
For example, Alf N Dor, 10th Jan, 15:25.
Now, it was nice seeing Peter mention it on his feed, & I'm hoping (for my longevity ) he laid it when I did. but my personal justification for laying it came about from a <100 sample (but the data was taken over a 7mth period)
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
80 is 80 whether you find them in a minute or a year. If something is rare you have to accept the outcome will be less certain.
But it depends what you're looking for though. If you're simply predicting up or down then 80 is a fair start I'd say, if you're looking for 1000/1 winners then 80 is utterly meaningless. Sample size should be proportional to what you are looking for, and if you want to find size of move, as well as up or down then 80 starts to feel a bit small. This stuff is probably week 1 of a stats degree, I think you should take a bullet for the team and find out http://www.open.ac.uk/courses/qualifications/q36
But it depends what you're looking for though. If you're simply predicting up or down then 80 is a fair start I'd say, if you're looking for 1000/1 winners then 80 is utterly meaningless. Sample size should be proportional to what you are looking for, and if you want to find size of move, as well as up or down then 80 starts to feel a bit small. This stuff is probably week 1 of a stats degree, I think you should take a bullet for the team and find out http://www.open.ac.uk/courses/qualifications/q36
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
There’s no doubt it’s risky, sure, but ; lets take the following seneraio:ShaunWhite wrote: ↑Mon Jan 14, 2019 6:41 am80 is 80 whether you find them in a minute or a year. If something is rare you have to accept the outcome will be less certain.
A trader could claim they have a single discretionary strategy, however a quant might label them as having, 100 minor objective strategies.
Assuming each edge had a profitable backtest over 100 markets (over the past yr); conventional analysis would deem each as inconclusive, and unlikely to succeed. Yet analysing the performance of the 100 as a single block = 10,000 sample. Suddenly we're into the realm of extreme validity, yet there's no overall difference
Now there's defo gonna be the curve fit argument, however conversely, 100 independent strategies technically lowers the chance of total failure, assuming a couple of the strategies do have some half decent logic , & aren't correlated.
Albeit, genuinely new trade ideas come at an absolute premium. Waiting for a bigger sample, far easier!!
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
Thread drift alert!!! Thread drift alert!!! Thread drift alert!!!
Not so much alert as a full scale incursion.
Not so much alert as a full scale incursion.
- ruthlessimon
- Posts: 2094
- Joined: Wed Mar 23, 2016 3:54 pm
Oh blimey... yeah. Memphis & Liam threw me off, I hadn't;t even read the OPShaunWhite wrote: ↑Mon Jan 14, 2019 8:03 pmThread drift alert!!! Thread drift alert!!! Thread drift alert!!!
Not so much alert as a full scale incursion.
I'll have to remember this post for the next sample related thread
- MemphisFlash
- Posts: 2128
- Joined: Fri May 16, 2014 10:12 pm
- Location: Leicester
not a bad afternoon on the dogs, while i was at work.
You do not have the required permissions to view the files attached to this post.
Hi, Can I get some advice for anybody? I would like to own profitable bot, on the start. I will be happy with 5€ prosit for one day. I wanted to sent message "MemphisFlash" one king of you, but I dont know how ? I dont see anywhere buttom for send message...
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
Markets are too small to just give away a bot. If he gives you one, his profit will go down. Everybody here makes their own and we don't even tell close friends, certainly not a complete stranger.
As for sending MemphisFlash a message asking for money, I hope you share his reply with us
- MemphisFlash
- Posts: 2128
- Joined: Fri May 16, 2014 10:12 pm
- Location: Leicester
OVER £1000 on the greyhounds alone today.
Snooker not to bad either.
You do not have the required permissions to view the files attached to this post.
Last edited by MemphisFlash on Tue Jan 15, 2019 10:09 pm, edited 1 time in total.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
If you are a trader, then look at your existing strategies, think about how they can be automated, read the BA user guide and try a few things.
Setting up automation is NOT programming, it's just learning how to use some fairly simple software.
Have you looked at the example automation files section of the forum for ideas?
ok, thank youShaunWhite wrote: ↑Tue Jan 15, 2019 9:15 pmIf you are a trader, then look at your existing strategies, think about how they can be automated, read the BA user guide and try a few things.
Setting up automation is NOT programming, it's just learning how to use some fairly simple software.
Have you looked at the example automation files section of the forum for ideas?